The study investigates temperature overshoot in a novel “current commitments” scenario that achieves large-scale negative emissions (of 3 different levels) in the 22^nd century and beyond. These scenarios would overshoot the 1.5C threshold for over 300 years. On top of these baseline scenarios the study implements SRM to keep temperatures to 1.5 C. They find, rather unsurprisingly, that in a scenario that exceeds 1.5C for centuries if SRM is deployed to keep temperatures below 1.5C then it would be deployed for centuries.

There is very little that this study would add to the literature. The study’s core finding is obvious, and the specific numerical value arrived at is determined by the scenario assumptions made by the authors and the one model that is applied. Furthermore, beyond showing the scenario(s) that the authors have created, the study has only 2 results: the length of time that SRM is deployed and the time-evolution of the cumulative carbon flux due to SRM. In my judgment there is not a sufficient depth of analysis or novelty in this work for it to be publishable in its current form.

Beyond the limited depth of analysis and lack of novelty, the results of this study are determined by the scenario assumptions made by the authors and by the insufficiently described ensemble of MAGICC6 model variants. While the scenario is fairly well described and quite reasonable, it is only 1 scenario (with 3 different CDR endings). There would be much more to analyse and discuss if a wider range of more and less ambitious scenarios were presented. The ensemble of MAGICC6 variants determines the results but there is insufficient description of how this ensemble is generated, nor is the reader given a quantitative assessment of its ECS and carbon cycle characteristics relative to more complex models or expert judgments.

In my judgment there is not a sufficient depth of analysis or novelty in this work for it to be publishable in its current form.

Review of Bauer et al. (2022)

The authors of this paper note in multiple places that little attention has been posed to the question of deployment timescales of SRM. Considering they cite other studies that have done similar analyses (and arrived to similar conclusions, but with far more deatil) one might wonder how many papers are enough before something has received more than “little attention”. Nevertheless, I agree that many times the words “temporary measures” or “stopgap measures” may give the impression that we’re talking about decadal deployments overall, whereas it is more likely that (were SRM ever implemented) the commitment would be longer than that, and definitely span more than one generation. And this has been discussed before.

However, when defining “how long” this “long” would be, one must rely on very long term assumptions about not just SRM, but climate policy and humanity in general. My fundamental issue with this paper is that the authors replicate partly what has been done in the extended RCPs and SSPs but without any of the carefulness employed by Meinhausen et al.: in the work around the extended scenarios, it is made clear multiple times that the goal is to look at the long-term Earth System response, and not to pretend to forecast how emissions (or CDR) will look like in 2250. It is a subtle distinction but an important one, especially if, as the authors do here, the simplistic assumptions around long term policies (i.e. “we extend the last 20 years of the century for 400 more”) are used to come up with rather precise numbers over the timescales of SRM deployment where the uncertainty is only related to climate sensitivity. But the work of, for instance, Lehner et al. (2020) clearly indicate that by the end of the century the main source of uncertainties in CMIP6 projection is the one related to scenarios - and indeed this is why the IPCC spans multiple ones. Selecting one arbitrary scenario, pretending it is valid for 500 years (I appreciate the validity of scenario-building, even on the very long-term, but imagine extrapolating the last 2 decades of the sixteenth century to find out how humanity would have fared in the year 2000) and then pretending that can give us a hard, numeric idea of SRM timescales of deployment just because MAGICC has been used to derive it seems incredibly weak to me. 

The authors admit the same at the beginning of Section 4: different assumptions over GHG emissions and CDR would alter the outcome considerably. But what the authors dismiss as obvious, it is not. I don’t see the merit of “determining” an outcome with a +/- 35 years precision that far into the future without including the context of other scenarios that span multiple sets of assumptions. Is a middle-of-the-road scenario that is currently tracking pledges the same scenario that would better track emissions at the end of the century? We can’t know, and therefore the only way we have is exploring multiple future pathways - being extremely clear that they are idealizations most likely to be wrong. But what the authors do is try to pretend their assumptions are the only “reasonable” ones, and fail to highlight enough how arbitrary their results are.

There are multiple parts of the manuscript where the authors make pretty arbitrary assumptions but try to pass them off as “neutral’. For instance:

“Specifically, we assume that the availability of SRM affects mitigation ambition and that after SRM initialization there is no incentive to increase ambition beyond the currently pledged targets. It is of course impossible to know how emissions would evolve under SRM”

If it’s impossible (and I agree) than how can one make that assumption so light-heartedly? Also, the authors say that unlike MacMartin et al. (2018) they consider mitigation, CDR and SRM in conjunction, but if you take an emission pathway that already exists and don’t modify it based on the presence of other components (SRM), you are indeed considering them as independent additive components. I would suggest you look into Drake et al. (2021) for an example of an exercise trying to consider these aspects in conjunction. The authors can also discuss some recent results relating to economic games where the concept of SRM is introduced, which generally point towards opposite results from what the authors assume (there are two that just recently came out, Talbot et al., 2022; Todd et al., 2022 and more in the past). If the authors chose to ignore these results, they need to explain why.

As another comment, in the abstract, the authors define their derived long times of deployment for SRM as a “risk”. But a risk compared to what? If it’s an abstract risk of just “doing” SRM (i.e. an almost ethical one: SRM is wrong and therefore the longer you do it, the more you are sinning) then it’s the authors personal view. If it’s a risk of negative effects (that the authors mention) and these risks trump those from climate change (which the authors don’t mention, plenty of literature around comparing risk of deployment versus risks of not deploying) then one might imagine that SRM wouldn’t hinder mitigation ambition but strengthen them.

In conclusion, I don’t think this manuscript is suitable for publication on ESD in its current form. The authors analyze their own scenario, based on their own assumptions, come up with a number and then consider and explain that number as an inevitable conclusion of any kind of SRM deployment under any possible scenario. What novel insight does that shed on anything? The analyses related to the carbon cycle are also to my eyes unremarkable, especially when other analyses based on more comprehensive models are already available. 

I agree with Reviewer 1 that this work would be far more robust if multiple scenarios where analyzed: if stronger mitigation and more CDR was available, as in other IPCC scenarios, how would these results look like? If the stabilization target was 2 instead of 1.5? This could probably be a way in which this work could become suitable for publication, together with a much more in depth discussion of methods used and a broader overview of past literature on the subject.

References

Drake, H. F., Rivest, R. L., Edelman, A., & Deutch, J. (2021). A simple model for assessing climate control trade-offs and responding to unanticipated climate outcomes. Environmental Research Letters, 16(10), 104012.

Lehner, F., Deser, C., Maher, N., Marotzke, J., Fischer, E. M., Brunner, L., Knutti, R., and Hawkins, E.: Partitioning climate projection uncertainty with multiple large ensembles and CMIP5/6, Earth Syst. Dynam., 11, 491–508, https://doi.org/10.5194/esd-11-491-2020, 2020.

Talbot M. Andrews, Andrew W. Delton, Reuben Kline, Anticipating moral hazard undermines climate mitigation in an experimental geoengineering game, Ecological Economics, Volume 196, 2022, 107421, ISSN 0921-8009, https://doi.org/10.1016/j.ecolecon.2022.107421.

Todd L. Cherry, Stephan Kroll, David M. McEvoy, David Campoverde & Juan Moreno-Cruz (2022) Climate cooperation in the shadow of solar geoengineering: an experimental investigation of the moral hazard conjecture, Environmental Politics, DOI: 10.1080/09644016.2022.2066285

General Comments:

They key point of the article is an important one: under a realistically pessimistic continuation of the new "business as usual" and feasible CDR rates,  keeping warming below 1.5ºC would require centuries of SRM before CDR could clean up the carbon mess. While I agree with RC1 that this point is obvious to experts, this does not mean that the obvious should go unstated! (RC2 somewhat agrees but finished by saying "this has been discussed before", without citing any specific examples.) I find the discussion in the literature unsatisfying and, at worst, deliberating misleading in a way that portrays SRM as less problematic than it really is.

However, I did find much the paper to be much longer than it needed to be; many paragraphs are unnecessary distractions from the key point. I also agree with RC1 and RC2 that the authors are trying to pass the scenarios as much too sophisticated or realistic. In contrast to RC1 and RC2, however, I would instead encourage the authors to dramatically simply their scenario and methods and submit it as a short "perspective", either in Earth System Dynamics or elsewhere. For example, they could use idealized emissions scenarios that are constant positive emissions —> linear decrease with prescribed slope —> constant negative emissions; these scenarios would be characterized by just 3 constants (the year emissions reductions begin; the emissions reductions rate; the CDR level) and the authors could briefly explore how these parameters affect the SRM timescale and how they relate to various scenarios discussed in the literature (including the somewhat realistically-pessimistic scenario they described here). While I do not think this article merits publication in Earth System Dynamics as is, I think that a revised article that is entirely (and only) about SRM timescale could merit publication.

Specific Comments:

I do not think the carbon cycle discussion merits inclusion in the article. The feedback is tiny compared to cumulative emissions and CDR, probably very model specific, and is not at all explained in terms of underlying processes.

RC2 mentioned that Drake et al. 2021 model scenarios with tradeoffs between mitigation, CDR, and SRM. Belaia et al. 2021 do as well (https://www.worldscientific.com/doi/full/10.1142/S2010007821500081). Neither explicitly state a timescale over which CDR has to be maintained, but it is obviously multiple centuries from their plots.

Given the discussion of climate changes on long timescale, it may be worth mentioning idealized scenario modelling that includes Sea Level Rise, e.g. Montero et al. (https://egusphere.copernicus.org/preprints/2022/egusphere-2022-135/).

Line 102: It should be explained where the 2.2 Wm^-2 value comes from.

Line 109: Is the 0.12 Wm^-2 relative to some long-time mean? What does this number signify?